1. Field of the Invention
Aspects of the present invention relate to computer modeling of environments. More specifically, aspects of the present invention relate to accurately capturing virtual environments.
2. Description of Related Art
A need exists for three-dimensional models of interior environments. From the gaming community to the military, actual renderings of building interiors based on actual buildings are being sought. For example, applications are being developed to provide training at the individual human level to facilitate improved military operations and capabilities as well as situational awareness. The training of individuals improves with a greater level of realism during the training exercise. For instance, benefits gained form mission rehearsal increase with the realism of the environment in which the mission is rehearsed.
Numerous applications exist for modeling outdoor terrain including satellite mapping and traditional surveying techniques. However, interior modeling techniques have been constrained to using pre-existing computer aided design drawings, building layout drawings, and a person's recollection of an environment. These modeling approaches may rely on out-of-date information (for example, blueprints of a 50-year-old building). Further, these modeling approaches fail to accurately define furniture types and furniture placement. In some situations, one may need to orient oneself based on the furniture and wall coverings of a room. Without this information, one may become lost and have to spend valuable time reorienting oneself with the environment. Further, one may want to know what is movable and what is not movable. Accordingly, realistic interiors are needed for simulation and training purposes.
Traditional methods exist for rendering building interiors including the use of computer aided design (CAD) drawings or laser scanning for the development of 3-D building models. However, as above, these approaches have not been helpful when attempting to interact a rendered environment in real-time. In some cases, the overload of information has made interaction with these environments difficult. For instance, original CAD drawings are not always current and laser scanning requires significant post processing. In particular, laser scanning about a point in space generates a cloud of data points which need to be processed to eliminate redundant points and convert them into easily-handled polygons. Further, laser scanning from one position is sub-optimal as points in a room are generally occluded by furniture. Blocking access to these points makes subsequent image processing difficult as a computer would need to determine whether non-planar points represent furniture or a curve in the wall of a room. To accurately determine the walls and contents of a room, multiple passes may be needed to attempt to eliminate blind spots caused by furniture and other objects. Further, these techniques do not capture data with knowledge of what is being captured.
Some techniques have been used to capture photographs of building interiors. For example, spherical photographic images have been used to capture photographic information surrounding a point in space in a room. While these photographic images may be assembled into a video stream representing a path through the room, deviation from the predefined path is not possible.
Accordingly, an improved system is needed for capturing and modeling interior spaces.